Professor Walsh makes an overwhelming case for separately analysing sub sections of the beverage alcohol market. In reply I shall comment on the relevance of his results to discussion of a policy to control consumption. Firstly, he suggests that his results establish such a positive connection between consumption (of spirits in particular) and real income that it would be very difficult to curb consumption in periods of rising real incomes. This conclusion is based on estimated coefficients of income two of which (beer, wine) are not statistically significant at the 10% level, whilst the third (spirits) is not significant at the 5% level. (See Table I). Whether this provides a sound basis for the conclusion is arguable. In McGuinness [2] I was reluctant to highlight the income effect precisely because the estimates of it were not statistically significant at conventional levels.' However, by using a quotation out of context Professor Walsh exaggerates my reluctance on this issue.2 In fact, in two of the three equations selected for particular focus in the later part of my paper, the estimated elasticities of alcohol consumption with respect to income are very close in magnitude to the figure given (0.45) by Professor Walsh. The issue is how far to proceed on the basis of estimates that are not statistically significant at conventional levels. In a forthcoming paper [3] my results offer stronger support for the existence of a significant effect of real income on alcohol consumption.